and of private capital are sunk, have been begun and have failed. The art of submarine telegraphic communication is by no means so far advanced as we had hoped, some little time ago indeed, owing to the destruction of several of the longest and most useful electric cables, it has, at this moment rather retrograded. The questions we are about to submit to our readers involve therefore not only facts of extreme interest, but certain problems which have yet to be fully solved. The evidence taken by the scientific Committee recently appointed to investigate the whole subject, is the first precise and authoritative account of it; and with these new and ample materials before us, we propose to trace the history of the more important cables which have been laid; to consider the causes of their failure; and then to discuss the position which the Government has assumed with reference to this species of commercial enterprise. The general principles upon which electric telegraphs, whether by land or sea, are constructed, are too well known to need repetition here. If a wire insulated from the earth be connected with the earth at one end, and with a battery communicating with the earth at the other, a current may be transmitted along the wire the strength of the current diminishing in an inverse ratio to the length of the wire. Or if both ends of a long wire be connected with the earth, currents will pass through the line apparently in consequence of variations in the electrical condition of the earth in different places- these are termed earth currents. Mr. Varley, the able electrician of the Electric and International Telegraph Company, observes that these currents are continually flowing about the earth, either in one direction or the other, throughout the day, and reach their maximum about 2.40 P.M. When magnetic storms or the aurora borealis occur, currents sufficiently powerful to interrupt the working of the lines flow sometimes in one direction, sometimes in the other, and often change from one direction to the other in the course of a few seconds. He has observed that there is no general line across England for these currents, but that the lines from London to King's Lynn and London to Southampton are frequently neutral, and the lines from London to Ipswich, London to Bristol, Hull to Manchester are powerfully affected. Professor Thomson gives in his evidence before the Government Telegraph Committee, the following account of a thunderstorm in Newfoundland being registered in Valentia, viz.:—

'On one occasion, whilst waiting for signals from Newfoundland, the mirror was found to be violently deflected at Valentia, so much so, that it had the appearance of being broken from its suspending thread; it turned out to be simply that the mirror was pressed for

cibly against the stop at the extreme end of its range. While I was looking into the mirror to ascertain whether there was any such accident, it suddenly turned round and went to the other side, there being no battery applied at all at the Valentia end. When communication was re-established, I asked what was wrong, and was told that a violent thunderstorm had been experienced at Newfoundland. "Great deflections and end put to earth for half an hour." This precaution having been very properly taken by the director of the station there to prevent the possibility of damage to the cable from lightning.'

In land lines the wires are generally insulated from the earth by being attached to supports made of some good insulating material, such as china or glass, which are fixed to the top of poles, twelve to fourteen feet high, placed from forty to sixty yards apart; the atmosphere being itself an admirable insulator especially when dry. But in crossing the sea it is necessary to cover the whole length of wire with an insulating material, and this insulating covering must itself be encased in a protecting sheath. Hence it will be seen that whilst for electric communication by land, only two elements, viz., the conducting wire and the insulators, are needed, the submarine telegraphic line must consist of three parts, viz.:-1st, the conducting wire; 2nd, the insulating covering; and 3rd, the protecting sheath.

Before describing these several component parts of a submarine cable, it is necessary, in order to elucidate the subject, to state the laws which govern the flow of electricity through insulated circuits of great length.

The conducting power of a conductor, as shown by Ohm, is in simple proportion to the area of its section and inversely to its length, when the quality of metal of the conductor is constant. The capacity of the insulated conductor for charge, or the electrostatic capacity, which influences most seriously the rate of signalling through it, depends on the ratio of the diameter of the insulator to the diameter of the conductor, and is independent of the absolute diameter of either; the facility for charging and discharging the cable is moreover proportioned to the square of the length, other things being the same. Professor

Wheatstone gives as a practical rule, that the induction varies directly with the length, and inversely with the square root of the diameter of the conductor and thickness of the insulating covering. Professor Thomson observes in his evidence given before the Committee on Submarine Telegraphs:

"The rate of signalling depends ultimately on the rapidity with which charge and discharge can be effected. I say ultimately; but before we reach this limit, there are many other considerations as regards

the sluggishness of the instruments, the system of more or less convenience for manipulation, and the susceptibility of the system for accuracy, all of which are to some extent uncertain. When these various circumstances are met in the most advantageous possible way, we come to a rate of speed in a line of 200 or 300 miles, which far exceeds the ordinary working rate. A machine could be got which could be worked through 200 or 300 miles at a very much greater rate than has yet been attained by any instrument in practical use. In estimating the speed of working through a long line we must consider the mere mechanical difficulties of very rapid action to be so completely overcome, as to give an extremely high speed in short lines; and from that basis proceed to estimate the rate of signalling through any length. If we could get three words a minute through 2000 miles, through 200 miles there would be 300 words a minute possible.'

The mechanical difficulties of manipulation, however, prevent this high speed being attained on the short lines, but on long lines when the retardation from induction is very great, the mechanical difficulties disappear, and the inductive difficulties limit the speed.

Another cause, which, curiously enough, limits the rate of speed is, that working a telegraph appears to cause nervous irritation in the clerks, and renders them prone to quarrel: if, for instance, one of the clerks carelessly sends a message indistinctly, the receiving clerk frequently gets out of temper, and serious delay results.

The conducting wire is of copper, and is usually made in a strand to diminish the chances of fracture to which a single wire is exposed. Copper is selected on account of the very superior conducting capacity of that metal, viz., seven times greater than that of iron. Copper wire is, however, deficient in strength, and it becomes permanently elongated when extended to a comparatively small amount. It has, therefore, been the practice to depend mainly for the strength of the cable on the protecting sheath. The insulating covering and the protecting sheath of a submarine cable, as generally made, possess more elasticity than copper wire. Consequently it has frequently happened that these after having been extended have returned to the original length, whilst the copper wire inside, which was equally extended, has remained permanently elongated, and has forced its way through its insulating covering of gutta percha. In order to prevent the undue extension of the copper wire, when a strain is brought on a cable, Mr. Allan, who has taken out several patents on the subject, proposes to place the strength of the cable close to the copper wire; for which purpose he covers it with fine steel wires, and covers these

again with the insulating material. The conducting power of steel is very low as compared with copper; hence the compound wire would have but little conducting capacity above that of the internal copper wire; but the induction, which varies with the area of the conductor, would be largely increased in a coated wire of this construction, as compared with a copper wire of the same conducting capacity, and this would necessitate a corresponding increase in the thickness of the insulating covering. It has been also suggested as possible, that in a considerable length of this compound wire an electric current might become divided, and that a portion would pass rapidly along the copper conductor, whilst the remainder would lag behind in the steel conductor, and that thus two currents would be exhibited at the opposite end. This objection could only be tested on a length of from 500 to 1000 miles of a cable so formed.

The conducting power of copper has been shown by Professor Thomson to vary with the purity of the metal, and it also changes with the temperature. The ordinary coppers of commerce are found to vary from pure copper as much as forty per cent. Indeed, in the case of copper wire a much greater variation has been found to exist, as is shown by experiments recently made by Mr. Mathiessen. If, for instance, the conducting power of pure copper be considered equal to 100, the conducting power of copper from Lake Superior, which contains traces of iron, silver and sub-oxide of copper, will be 92.5; that of copper from the Burra Burra mines in Australia, which contains traces of iron and sub-oxide of copper, will be 88; whilst that of Russian (Demidoff) copper, which contains arsenic, iron, nickel, and sub-oxide of copper, will be 59; and that of Spanish (Rio Tinto) copper, which contains two per cent. of arsenic, traces of lead, iron, nickel, and sub-oxide of copper, will be only 14, and is thus of lower conducting power than iron. The presence of suboxide of copper is especially injurious to the conducting power of copper, and the presence of the metalloids is as a rule more injurious than that of foreign metals. Mr. Mathiessen's valuable report on this subject, concludes by showing that no substance added to copper increases its conducting power, and that the purest obtainable copper should therefore be used in the manufacture of submarine cables. To secure this, the contract for the cable should specify a given resistance per knot, in which case every failure in quality would have to be compensated, at the manufacturer's expense, by extra thickness. Copper is not, however, a good metal to employ as a standard, because it oxydises easily, and the conducting power varies much with the temperature.

The material which has hitherto been almost exclusively used for the insulating covering is gutta percha. This substance is a good non-conductor of electricity, and from its viscous character when warmed it adheres easily to a wire. In order to coat the wires, the gutta percha is forced out of a circular die, through the centre of which the wire is passed, and draws away with it the gutta percha covering thus forced on to it. A compound made of Stockholm tar, resin, and gutta percha, is placed on the wire before the gutta percha covering is placed on it, and also between each coating of gutta percha. In placing the gutta percha on the wire, the causes of injury to be guarded against are: 1st, air bubbles; 2nd, the eccentricity of the wire, in which a thinner layer of gutta percha exists between the wire and the surface, than at other parts; 3rd, porosity of the gutta percha; 4th; the presence of foreign bodies which connect the copper wire inside the gutta percha with the water outside; 5th, bad joints at the places where different lengths of wire are joined together; 6th, small punctures. Any of these permit the electric current flowing through a wire to pass through the gutta percha at the place where the injury occurs more easily than at other places, and the passage of the electricity generated by strong battery power, produces a chemical action which gradually destroys the gutta percha and exposes the copper wire, until in turn it also becomes eaten away. Mr. Jenkin in one of his papers observes:-

Accident put me in possession of a fault caused by an air bubble, which, I think, throws light on many recent failures of cables which tested well when first laid down. I traced it, until I ascertained its position in the gutta percha to within one inch. There were signs that the gutta percha had been heated during manufacture. Nevertheless I was unable to find any visible flaw in the gutta percha. A little white speck looked suspicious, but on wiping away I could see no hole. During the tests (in fresh water) the fault got a little worse, but not much except perhaps for an instant on the first admission of a negative current. The current was continually reversed during the tests. I put the fault into a wineglass of salt water and tested it once more (with a negative current). I at once saw a little row of bubbles rise from the spot where the white mark had been. I took the fault out of the water almost immediately, and a little hole could now be distinctly seen. I replaced the fault in the wineglass, putting the bulb of a thermometer close to the little hole. Bubbles escaped rapidly from the fault, the mercury in the thermometer rose 5°, and in three minutes the fault had lost almost all resistance. A positive current as well as a negative current was used. The hole was now th in. long, and th broad. A hollow extending to some distance on each side, indicated the presence of an air bubble. A second hole had begun